MVOS_HSI: A Python Library for Preprocessing Agricultural Crop Hyperspectral Data

A research team including Rishik Aggarwal, Krisha Joshi, and USDA scientists has released MVOS_HSI, an open-source Python library designed to solve a critical bottleneck in agricultural AI research. Hyperspectral imaging (HSI) captures hundreds of spectral bands per pixel, revealing biochemical details about plant health invisible to standard cameras, but processing this complex data has traditionally relied on disorganized, lab-specific MATLAB or Python scripts. MVOS_HSI consolidates these fragmented workflows into a single, standardized package that researchers can run as an importable library or from the command line.

The library provides an end-to-end pipeline specifically for leaf-level HSI data. It starts by calibrating raw data from common ENVI file formats, then automatically detects and clips individual leaves using multiple vegetation indices including NDVI, CIRedEdge, and GCI. For machine learning applications, MVOS_HSI includes built-in data augmentation tools to create training-time variations, plus utilities for visualizing spectral profiles. By packaging these common preprocessing steps, the tool addresses reproducibility challenges in plant phenotyping, allowing different research teams to generate consistent, comparable results from hyperspectral data.